/* Copyright 2003, Carnegie Mellon, All Rights Reserved */
package edu.cmu.minorthird.classify;
import java.awt.GridBagConstraints;
import java.awt.GridBagLayout;
import java.awt.event.ActionEvent;
import java.io.ByteArrayOutputStream;
import java.io.PrintStream;
import java.util.Iterator;
import javax.swing.AbstractAction;
import javax.swing.JButton;
import javax.swing.JComponent;
import javax.swing.JLabel;
import javax.swing.JPanel;
import javax.swing.JProgressBar;
import javax.swing.border.TitledBorder;
import org.apache.log4j.Logger;
import edu.cmu.minorthird.classify.ClassifyCommandLineUtil.MultiTrainTestParams;
import edu.cmu.minorthird.classify.ClassifyCommandLineUtil.SeqTrainTestParams;
import edu.cmu.minorthird.classify.algorithms.knn.KnnLearner;
import edu.cmu.minorthird.classify.algorithms.linear.MaxEntLearner;
import edu.cmu.minorthird.classify.algorithms.linear.NaiveBayes;
import edu.cmu.minorthird.classify.algorithms.linear.VotedPerceptron;
import edu.cmu.minorthird.classify.algorithms.svm.SVMLearner;
import edu.cmu.minorthird.classify.algorithms.trees.AdaBoost;
import edu.cmu.minorthird.classify.algorithms.trees.DecisionTreeLearner;
import edu.cmu.minorthird.classify.experiments.CrossValSplitter;
import edu.cmu.minorthird.classify.experiments.CrossValidatedDataset;
import edu.cmu.minorthird.classify.experiments.Evaluation;
import edu.cmu.minorthird.classify.experiments.RandomSplitter;
import edu.cmu.minorthird.classify.experiments.StratifiedCrossValSplitter;
import edu.cmu.minorthird.classify.experiments.Tester;
import edu.cmu.minorthird.classify.multi.InstanceFromPrediction;
import edu.cmu.minorthird.classify.multi.MultiClassLabel;
import edu.cmu.minorthird.classify.multi.MultiClassifier;
import edu.cmu.minorthird.classify.multi.MultiCrossValidatedDataset;
import edu.cmu.minorthird.classify.multi.MultiDataset;
import edu.cmu.minorthird.classify.multi.MultiEvaluation;
import edu.cmu.minorthird.classify.multi.MultiExample;
import edu.cmu.minorthird.classify.sequential.CollinsPerceptronLearner;
import edu.cmu.minorthird.classify.sequential.CrossValidatedSequenceDataset;
import edu.cmu.minorthird.classify.sequential.GenericCollinsLearner;
import edu.cmu.minorthird.classify.sequential.SequenceDataset;
import edu.cmu.minorthird.classify.transform.FrequencyBasedTransformLearner;
import edu.cmu.minorthird.classify.transform.InfoGainTransformLearner2;
import edu.cmu.minorthird.classify.transform.T1InstanceTransformLearner;
import edu.cmu.minorthird.classify.transform.TFIDFTransformLearner;
import edu.cmu.minorthird.classify.transform.TransformingBatchLearner;
import edu.cmu.minorthird.util.BasicCommandLineProcessor;
import edu.cmu.minorthird.util.CommandLineProcessor;
import edu.cmu.minorthird.util.IOUtil;
import edu.cmu.minorthird.util.JointCommandLineProcessor;
import edu.cmu.minorthird.util.ProgressCounter;
import edu.cmu.minorthird.util.StringUtil;
import edu.cmu.minorthird.util.gui.ComponentViewer;
import edu.cmu.minorthird.util.gui.Console;
import edu.cmu.minorthird.util.gui.SmartVanillaViewer;
import edu.cmu.minorthird.util.gui.TypeSelector;
import edu.cmu.minorthird.util.gui.Viewer;
import edu.cmu.minorthird.util.gui.ViewerFrame;
/**
* Main UI program for the 'classify' package.
*
* @author William Cohen
*/
public class TrainTest{
private static Logger log=Logger.getLogger(UI.class);
private static final Class<?>[] SELECTABLE_TYPES=
new Class[]{
DataClassificationTask.class,
ClassifyCommandLineUtil.SimpleTrainTestParams.class,
ClassifyCommandLineUtil.MultiTrainTestParams.class,
ClassifyCommandLineUtil.SeqTrainTestParams.class, //ClassifyCommandLineUtil.TrainTestParams.class,
ClassifyCommandLineUtil.Learner.SequentialLearner.class,
ClassifyCommandLineUtil.Learner.ClassifierLearner.class,KnnLearner.class,
NaiveBayes.class,VotedPerceptron.class,SVMLearner.class,
DecisionTreeLearner.class,AdaBoost.class,BatchVersion.class,
TransformingBatchLearner.class,
MaxEntLearner.class,
// transformations
FrequencyBasedTransformLearner.class,InfoGainTransformLearner2.class,
T1InstanceTransformLearner.class,TFIDFTransformLearner.class,
// sequential learner
CollinsPerceptronLearner.class,GenericCollinsLearner.class,
// splitters
CrossValSplitter.class,RandomSplitter.class,
StratifiedCrossValSplitter.class,};
// private static final Set LEGAL_OPS=
// new HashSet(Arrays.asList(new String[]{"train","test","trainTest"}));
public static class DataClassificationTask implements
CommandLineProcessor.Configurable,/*Saveable,*/Console.Task{
private ClassifyCommandLineUtil.TrainTestParams trainTestParams=
new ClassifyCommandLineUtil.TrainTestParams();
public Object resultToShow;
public boolean useGUI;
//public String type;
public Console.Task main;
// for gui
public ClassifyCommandLineUtil.TrainTestParams getTrainTestParameters(){
return trainTestParams;
}
public void setTrainTestParameters(ClassifyCommandLineUtil.TrainTestParams p){
trainTestParams=p;
}
public String getTrainTestParamsHelp(){
return "Define what type of experiment you would like to run: <br>"
+"Simple - Standard classify experiment <br> "
+"Multi - Classify Experiment with Multiple labels per example <br>"
+"Seq - Classify experiment with a Sequential Dataset, where each example has a history, <br> "
+" and uses a Sequential Learner";
}
protected class GUIParams extends BasicCommandLineProcessor{
//public void type(String s) {type=s;}
public void gui(){
useGUI=true;
if(ClassifyCommandLineUtil.TrainTestParams.type!=null)
trainTestParams=ClassifyCommandLineUtil.TrainTestParams.type;
else
trainTestParams=new ClassifyCommandLineUtil.SimpleTrainTestParams();
}
@Override
public void usage(){
System.out.println("presentation parameters:");
System.out
.println(" -gui use graphic interface to set parameters");
System.out.println();
}
}
public String getDatasetFilename(){
return trainTestParams.trainDataFilename;
}
@Override
public CommandLineProcessor getCLP(){
JointCommandLineProcessor jlpTrainTest=
new JointCommandLineProcessor(new CommandLineProcessor[]{
new GUIParams(),trainTestParams});
return jlpTrainTest;
}
/** Returns whether base.labels exits */
@Override
public boolean getLabels(){
return(getDatasetFilename()!=null);
}
public MultiDataset annotateData(MultiClassifier multiClassifier,
MultiDataset md){
MultiDataset annotatedDataset=new MultiDataset();
for(Iterator<MultiExample> i=md.multiIterator();i.hasNext();){
MultiExample ex=i.next();
Instance instance=ex.asInstance();
MultiClassLabel predicted=
multiClassifier.multiLabelClassification(instance);
Instance annotatedInstance=
new InstanceFromPrediction(instance,predicted.bestClassName());
MultiExample newEx=
new MultiExample(annotatedInstance,ex.getMultiLabel(),ex
.getWeight());
annotatedDataset.addMulti(newEx);
}
return annotatedDataset;
}
// main action
@Override
public void doMain(){
//Work in more tests
if(trainTestParams.getTrainData()==null){
System.out
.println("The training data needs to be specified with the -data option.");
return;
}
if(trainTestParams instanceof ClassifyCommandLineUtil.SeqTrainTestParams&&
(!(trainTestParams.getTrainData() instanceof SequenceDataset))){
System.out.println("The training data should be a sequence dataset");
return;
}
if(trainTestParams.multi>0&&
(!(trainTestParams.getTrainData() instanceof MultiDataset))){
System.out.println("The training data should be a multi dataset");
return;
}
if(trainTestParams.showData)
new ViewerFrame("Training data",trainTestParams.getTrainData().toGUI());
if(trainTestParams.showTestDetails&&trainTestParams.typeString.equals("seq")){
CrossValidatedSequenceDataset cvd=
new CrossValidatedSequenceDataset(
trainTestParams.seqLnr.seqLearner,
(SequenceDataset)trainTestParams.getTrainData(),
((SeqTrainTestParams)trainTestParams).splitter);
trainTestParams.resultToShow=cvd;
trainTestParams.resultToSave=cvd.getEvaluation();
((Evaluation)trainTestParams.resultToSave).summarize();
}else if(!trainTestParams.showTestDetails&&trainTestParams.typeString.equals("seq")){
Evaluation e=
Tester.evaluate(trainTestParams.seqLnr.seqLearner,
(SequenceDataset)trainTestParams.getTrainData(),
trainTestParams.sequenceSplitter);
trainTestParams.resultToShow=trainTestParams.resultToSave=e;
((Evaluation)trainTestParams.resultToSave).summarize();
}else if(trainTestParams.showTestDetails&&
trainTestParams.typeString.equals("multi")){
MultiCrossValidatedDataset cvd=
new MultiCrossValidatedDataset(trainTestParams.clsLnr.clsLearner,
(MultiDataset)trainTestParams.getTrainData(),
((MultiTrainTestParams)trainTestParams).splitter);
trainTestParams.resultToShow=cvd;
trainTestParams.resultToSave=cvd.getEvaluation();
((MultiEvaluation)trainTestParams.resultToSave).summarize();
}else if(!trainTestParams.showTestDetails&&
trainTestParams.typeString.equals("multi")){
MultiEvaluation e=
Tester.multiEvaluate(trainTestParams.clsLnr.clsLearner,
(MultiDataset)trainTestParams.getTrainData(),
((MultiTrainTestParams)trainTestParams).splitter);
trainTestParams.resultToShow=trainTestParams.resultToSave=e;
((MultiEvaluation)trainTestParams.resultToSave).summarize();
}else if(trainTestParams.showTestDetails&&
!(trainTestParams.equals("seq"))){
CrossValidatedDataset cvd=
new CrossValidatedDataset(trainTestParams.clsLnr.clsLearner,
trainTestParams.getTrainData(),trainTestParams.splitter);
trainTestParams.resultToShow=cvd;
trainTestParams.resultToSave=cvd.getEvaluation();
((Evaluation)trainTestParams.resultToSave).summarize();
}else if(!trainTestParams.showTestDetails&&
!(trainTestParams.typeString.equals("seq"))){
Evaluation e=
Tester.evaluate(trainTestParams.clsLnr.clsLearner,trainTestParams
.getTrainData(),trainTestParams.splitter);
trainTestParams.resultToShow=trainTestParams.resultToSave=e;
((Evaluation)trainTestParams.resultToSave).summarize();
}
resultToShow=trainTestParams.resultToShow;
// attach all the command-line arguments to the resultToSave, as properties
/*for (Iterator i=clp.propertyList().iterator(); i.hasNext(); ) {
String prop = (String)i.next();
((Evaluation)resultToSave).setProperty(prop,clp.propertyValue(prop));
}*/
if(trainTestParams.showResult)
new ViewerFrame("Result",new SmartVanillaViewer(
trainTestParams.resultToShow));
if(trainTestParams.saveAs!=null){
if(IOUtil.saveSomehow(trainTestParams.resultToSave,
trainTestParams.saveAs)){
log.info("Result saved in "+trainTestParams.saveAs);
}else{
log.error("Can't save "+trainTestParams.resultToSave.getClass()+
" to "+trainTestParams.saveAs);
}
}
}
@Override
public Object getMainResult(){
return resultToShow;
}
//
// implements Saveable
//
/*public String[] getFormatNames() { return clp.getFormatNames(); }
public String getExtensionFor(String format) { return clp.getExtensionFor(format); }
public void saveAs(File file, String format) throws IOException { clp.saveAs(file,format); }
public Object restore(File file) throws IOException
{
DataClassificationTask task = new DataClassificationTask();
task.clp.config(file.getAbsolutePath());
return task;
}*/
// gui around main action
public void callMain(final String[] args){
try{
getCLP().processArguments(args);
if(!useGUI){
doMain();
}else{
main=this;
final Viewer v=new ComponentViewer(){
static final long serialVersionUID=20080128L;
@Override
public JComponent componentFor(Object o){
Viewer ts=
new TypeSelector(SELECTABLE_TYPES,"selectableTypes.txt",
DataClassificationTask.class);
ts.setContent(o);
// we'll put the type selector in a nice panel
JPanel panel=new JPanel();
panel.setBorder(new TitledBorder(StringUtil.toString(args,
"Command line: ",""," ")));
panel.setLayout(new GridBagLayout());
GridBagConstraints gbc;
// another panel to allow parameter modifications
JPanel subpanel1=new JPanel();
subpanel1.setBorder(new TitledBorder("Parameter modification"));
//subpanel1.add(new JLabel("Use the edit button to change the parameters given in the command line"));
subpanel1.add(ts);
gbc=Viewer.fillerGBC();
gbc.weighty=0;
panel.add(subpanel1,gbc);
// a control panel for controls
JPanel subpanel2=new JPanel();
subpanel2.setBorder(new TitledBorder("Execution controls"));
// a button to show the results
final JButton viewButton=
new JButton(new AbstractAction("View results"){
static final long serialVersionUID=20080128L;
@Override
public void actionPerformed(ActionEvent event){
Viewer rv=new SmartVanillaViewer();
rv.setContent(getMainResult());
new ViewerFrame("Result",rv);
}
});
viewButton.setEnabled(false);
// another panel for error messages and other outputs
JPanel errorPanel=new JPanel();
errorPanel
.setBorder(new TitledBorder("Error messages and output"));
final Console console=
new Console(main,getDatasetFilename()!=null,viewButton);
errorPanel.add(console.getMainComponent());
// a button to start this thread
JButton goButton=new JButton(new AbstractAction("Start task"){
static final long serialVersionUID=20080128L;
@Override
public void actionPerformed(ActionEvent event){
console.start();
}
});
// and a button to show the current labels
JButton showLabelsButton=
new JButton(new AbstractAction("Show train data"){
static final long serialVersionUID=20080128L;
@Override
public void actionPerformed(ActionEvent ev){
new ViewerFrame(
"Labeled TextBase",
new SmartVanillaViewer(trainTestParams.getTrainData()));
}
});
// and a button to clear the errorArea
JButton clearButton=
new JButton(new AbstractAction("Clear window"){
static final long serialVersionUID=20080128L;
@Override
public void actionPerformed(ActionEvent ev){
console.clear();
}
});
// and a button for help
JButton helpParamsButton=
new JButton(new AbstractAction("Parameters"){
static final long serialVersionUID=20080128L;
@Override
public void actionPerformed(ActionEvent ev){
PrintStream oldSystemOut=System.out;
ByteArrayOutputStream outBuffer=
new ByteArrayOutputStream();
System.setOut(new PrintStream(outBuffer));
//clp.usage();
console.append(outBuffer.toString());
System.setOut(oldSystemOut);
}
});
subpanel2.add(goButton);
subpanel2.add(viewButton);
subpanel2.add(showLabelsButton);
subpanel2.add(clearButton);
subpanel2.add(new JLabel("Help:"));
subpanel2.add(helpParamsButton);
gbc=Viewer.fillerGBC();
gbc.weighty=0;
gbc.gridy=1;
panel.add(subpanel2,gbc);
gbc=Viewer.fillerGBC();
gbc.weighty=1;
gbc.gridy=2;
panel.add(errorPanel,gbc);
// now some progress bars
JProgressBar progressBar1=new JProgressBar();
JProgressBar progressBar2=new JProgressBar();
JProgressBar progressBar3=new JProgressBar();
ProgressCounter.setGraphicContext(new JProgressBar[]{
progressBar1,progressBar2,progressBar3});
gbc=Viewer.fillerGBC();
gbc.weighty=0;
gbc.gridy=3;
panel.add(progressBar1,gbc);
gbc=Viewer.fillerGBC();
gbc.weighty=0;
gbc.gridy=4;
panel.add(progressBar2,gbc);
gbc=Viewer.fillerGBC();
gbc.weighty=0;
gbc.gridy=5;
panel.add(progressBar3,gbc);
return panel;
}
};
v.setContent(this);
String className=
this.getClass().toString().substring("class ".length());
new ViewerFrame(className,v);
}
}catch(Exception e){
e.printStackTrace();
System.out.println("Use option -help for help");
}
}
}
public static void main(String[] args){
new DataClassificationTask().callMain(args);
}
}